9/04/2019

Aloha

The spread of zebra mussels

The Minnesota invasion

Data from USGS Nonindigenous Aquatic Species Information Resource

Impacts of the mussel invasion


image: Mussel Prevention Program, San Luis Obispo County

  • Changes in water chemistry
  • Decreases in plankton densities
    • Decreases in native mussel populations
    • Changes in water clarity
    • Increases in plant cover
  • Economic impacts on power industry

Managing the invasion

Surveying mussels: Year 1

Year 1 field crew

image: Naomi Blinick

image: Naomi Blinick

Why shouldn’t we just go out and start counting?

image: books.google.com

image: books.google.com

Existing survey designs


image: Naomi Blinick


  • Discovery in low densities (timed searches)
  • High-densities (quadrat surveys)



But what about designs for low to moderate densities?

Transect sampling is one approach to cover area quickly

Distance sampling is one approach to cover area quickly and account for imperfect detection

Extra information yields detection estimates

Leca, J., N. Gunst, A. Rompis, G. Soma, I. G. A. Arta Putra, and I. N. Wandia (2013) Population Density and Abundance of Ebony Leaf Monkeys (Trachypithecus auratus) in West Bali National Park, Indonesia, Primate Conservation 26(1), 133-144.


Important assumptions in conventional distance sampling


  • Density away from transect line is homogenous
  • Detection on the transect line is perfect
  • Animals do not move before detection
  • Measurements are exact

First lake: Lake Sylvia

Note the hump shape, are we detecting everything on the transect line?

We needed to add a second observer!

  • Adds a mark-recap component to the distance survey.
    • First diver counts, followed by second diver
  • Determining which mussels are detected by one or both observers allows us to estimate density on the transect line.

Second lake: Lake Burgan

We can now estimate detection probabilities

image: Naomi Blinick

image: Naomi Blinick


Estimated density without detection \(0.08\) mussels/m\(^2\).

Estimated density without detection \(0.25\) \((0.07)\) mussels/m\(^2\) .

Lessons from Year 1

  • Detection on the transect line is far from perfect
  • Need to use double-observer surveys
    • Significant heterogeneity between observers
  • Do not need to stratify effort
  • Dive surveys are hard
image: Aislyn Keyes

image: Aislyn Keyes

Surveying mussels: Year 2

Dive team

But how do distance surveys compare to quadrat surveys?

Last year we demonstrated that distance surveying is possible, but (when) is it preferable?

image: Jake Ferguson

image: Jake Ferguson

Given a fixed amount of time, which method performs best?

Two sources of variance: counts and detection

Compare designs across densities


Results

Lessons from Year 2

  • Distance sampling is preferable at low to moderate densities


image: Aislyn Keyes

image: Aislyn Keyes

Generalizing these results

Analogy to Optimal Foraging Theory


From Charnov, E. L. 1976. Optimal foraging: the marginal value theorem. Theoretical Population Biology 9:129–136

From Charnov, E. L. 1976. Optimal foraging: the marginal value theorem. Theoretical Population Biology 9:129–136

Break up the survey into steps, determine the time it takes to complete each step

We can use this to predict the optimal strategy

Predicts optimum changes at a lower density than we observed empirically!

Implications of these results

  • Treating the surveyor as a predator we can find an optimal strategy
  • The surveyor speed may be ‘tuned’ to the density
  • Leads to simpler variance formula than traditional adaptive cluster sampling

Acknowledgements

jakeferg@hawaii.edu

John Fieberg

Michael McCartney

Naomi Blinick

Leslie Schroeder

Sarah Baker

Aislyn Keyes

Austin Hilding

Other materials

Quadrat sampling error